How a Google DeepMind Spinoff Hunts Hidden Drug Targets

2026-06-11 · Source: IEEE Spectrum · Field: Science & Research — Life Sciences & Biology, Health & Medical Research, Artificial Intelligence & Machine Learning · Depth: Advanced, short

Summary

Isomorphic Labs, a Google DeepMind spinout, has developed the Isomorphic Drug Design Engine (IsoDDE), a unified computational system aimed at accelerating drug discovery. Building on AlphaFold's protein structure prediction, IsoDDE addresses its limitations by predicting how proteins and drug molecules interact, including structure prediction, pocket identification, and binding affinity. The company, which recently raised US \$2.1 billion and partnered with Novartis and Eli Lilly, demonstrated IsoDDE's capability by accurately predicting a newly discovered "cryptic pocket" on the cereblon protein. This type of pocket is non-obvious in the unbound state and only appears with specific ligand binding. IsoDDE expands the toolkit for tackling diseases by making more protein targets tractable, applicable not only to small molecules but also to antibodies, molecular glues, and peptides, moving beyond the misconception that structure modeling alone solves drug discovery.

Key takeaway

For AI Scientists and Machine Learning Engineers focused on drug discovery, you should evaluate unified computational systems like IsoDDE to overcome limitations of structure-only prediction models. This approach enables the identification of novel, cryptic protein binding pockets and expands the range of druggable targets for various therapeutic modalities, accelerating the development of new medicines. Consider integrating such multi-endpoint platforms to enhance your drug design workflows.

Key insights

IsoDDE enables discovery of novel, cryptic protein binding pockets, expanding druggable targets beyond AlphaFold's capabilities.

Principles

• Protein folding models alone are insufficient for drug design.
• Novel drug mechanisms require models generalizing to distant regions.
• Unified systems with multiple endpoints are crucial.

Method

IsoDDE is a unified computational system predicting protein-ligand interactions through structure prediction, pocket identification, and binding affinity prediction, validated by finding cryptic pockets.

In practice

• Identify cryptic pockets on disease-associated proteins.
• Predict binding of small molecules, antibodies, and peptides.
• Validate novel binding sites with high accuracy.

Topics

• Drug Discovery
• Isomorphic Labs
• IsoDDE
• Protein Structure Prediction
• Cryptic Pockets
• AlphaFold
• Machine Learning in Pharma

Best for: Investor, AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• ‘An AlphaFold 4’ – scientists marvel at DeepMind drug spin-off’s exclusive new AI — Isomorphic Labs' proprietary IsoDDE model significantly advances AI drug discovery, outperforming open-source alternatives in binding affinity prediction.
• Why We're Not Close to AI Drugs Yet | Isomorphic Labs — Drug design comprises multiple distinct ML problems, not a single one, despite AlphaFold 2's protein structure prediction success.
• Probe Before You Edit: Probing-Guided Molecular Optimization for LLM Agents in Structure-Based Drug Design — Probing molecular edit responses guides LLM agents to jointly optimize binding affinity and druggability in drug design.
• Bridging the phenotype-target gap for molecular generation via multi-objective reinforcement learning — ExMolRL combines phenotype-guided generation with target-aware reinforcement learning for de novo drug discovery.
• 😺 🎙️ Watch: How Isomorphic Labs works to drug "undruggable" diseases — AI, particularly AlphaFold-derived technology, is dramatically accelerating drug discovery and tackling previously "undruggable" targets.
• AF2BIND: predicting small-molecule binding sites using the pair representation of AlphaFold2 — AlphaFold2's internal representations can be repurposed to accurately predict de novo small-molecule binding sites.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by IEEE Spectrum.